Biological Journal of the Linnean Society, 2008, 93, 621640.

With 6 gures

Molecular phylogeny and diversication history of Prosopis (Fabaceae: Mimosoideae)

SANTIAGO ANDRS CATALANO1*, JUAN CSAR VILARDI1, DANIELA TOSTO1,2 and BEATRIZ OFELIA SAIDMAN1
1

Departamento de Ecologa Gentica y Evolucin, Facultad de Ciencias Exactas y Naturales, Universidad Nacional de Buenos Aires. Intendente Giraldes 2160, Ciudad Universitaria, C1428EGA - Capital Federal, Argentina. 2 Instituto de Biotecnologa, CICVyA INTA Castelar, CC 25 Castelar 1712, Argentina
Received 29 December 2006; accepted for publication 31 May 2007

The genus Prosopis is an important member of arid and semiarid environments around the world. To study Prosopis diversication and evolution, a combined approach including molecular phylogeny, molecular dating, and character optimization analysis was applied. Phylogenetic relationships were inferred from ve different molecular markers (matK-trnK, trnL-trnF, trnS-psbC, G3pdh, NIA). Taxon sampling involved a total of 30 Prosopis species that represented all Sections and Series and the complete geographical range of the genus. The results suggest that Prosopis is not a natural group. Molecular dating analysis indicates that the divergence between Section Strombocarpa and Section Algarobia plus Section Monilicarpa occurred in the Oligocene, contrasting with a much recent diversication (Late Miocene) within each of these groups. The diversication of the group formed by species of Series Chilenses, Pallidae, and Ruscifoliae is inferred to have started in the Pliocene, showing a high diversication rate. The moment of diversication within the major lineages of American species of Prosopis is coincident with the spreading of arid areas in the Americas, suggesting a climatic control for diversication of the group. Optimization of habitat parameters suggests an ancient occupation of arid environments by Prosopis species. 2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640.

ADDITIONAL KEYWORDS: ancestral environments aridication character optimization molecular

dating penalized likelihood.

INTRODUCTION
The genus Prosopis L. (Fabaceae) has approximately 45 species, which are important members of arid and semiarid environments (Fig. 1). Some of these species are characteristic of the driest regions in the world. For example, Prosopis tamarugo F. Philippi, is one of the few tree species capable of surviving in the extremely arid climate of the Atacama desert in Northern Chile. Other species are distinctive of the large deserts of North America (Tamaulipas, Sonora, Chiuahua) and of arid and semiarid regions of South America (Monte, Patagonia, Puna, and Chaco). However, only a few representatives of the genus,
*Corresponding author. E-mail: sacatalano@gmail.com

such as Prosopis africana (Guill., Perr., & Rich.) Taubert, are partially distributed in subhumid tropical or subtropical regions. From a taxonomic point of view, the genus is divided into ve Sections, based mainly on the presence and type of armature and shoot structure (Burkart, 1976). The two main Sections (Algarobia and Strombocarpa) are native to North and South America and include approximately 90% of all the species of the genus. Section Monilicarpa comprises only one species endemic to central Argentina. Sections Anonychium and Prosopis have exclusively Old World representatives. Prosopis africana, the only species of Section Anonychium, is native to the Soudano-Guinean zone and neighbouring areas of Africa, from Senegal in the west to Sudan and Kenya in the east (Pasiecznik et al., 2001).

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640

621

622

S. A. CATALANO ET AL.

Figure 1. Natural distribution of Prosopis after Pasiecznik et al. (2001). Number of native species are indicated for each region.

Section Prosopis comprises three species that are native to North Africa and Asia, stretching east to India, north to Georgia and Turkmenistan, and west to Algeria along the North African coast (Pasiecznik et al., 2001). Different studies have evaluated the relationships among Prosopis species (Saidman & Vilardi, 1987, 1993; Saidman et al., 1996, 1998; Ramrez et al., 1999; Bessega, Saidman & Vilardi, 2005; Bessega, Vilardi & Saidman, 2006). However, these generally involved phenetic analyses and included species of only some of the Series and Sections of the genus. In addition, none of these studies have evaluated the nature of the Prosopis generic limits as either outgroups were not included (Ramrez et al., 1999; Bessega et al., 2005), or these were distantly related to Prosopis species (Bessega et al., 2006). The only cladistic analysis performed (Bessega et al., 2006) suggested that most of the groups taxonomically recognized are not monophyletic. Several ideas were proposed about the timing of Prosopis diversication (Pasiecznik et al., 2001). Burkart (1976), based on the presence of the genus in both New and Old World, suggested a late Cretaceous or Early Tertiary origin for the genus, previous to the formation of the Atlantic Ocean. Following this idea, Ramrez et al. (1999) suggested that the genus would have originated 130 Mya. However, this time frame is strongly contradicted by both palaeonthological evidence (Herendeen, Crepet & Dilcher, 1992) and molecular dating (Lavin, Herendeen &

Wojciechowski, 2005) because both support the notion that the subfamily Mimosoideae originated between 4250 Mya. Burkart & Simpson (1977) also considered that Prosopis is an old genus that diverged early into several principal lineages but that, within some of these lineages, more recent episodes of expansion and isolation produced further speciation. To date, no attempt has been made to evaluate the divergence times within Prosopis using molecular dating. The observation of hybrid formation between some species of Section Algarobia, together with their high morphological similarity and partially overlapping geographical distributions, led to the consideration that this group is a syngameon (Palacios & Bravo, 1981). However, recent molecular studies have supported the biological meaning of the specic limits within this group (Saidman & Vilardi, 1987; Burghardt, 1995; Bessega et al., 2000) and indicate the possible existence of isolating mechanisms that restrict the introgression between its species. Previous cladistic analysis (Bessega et al., 2006) suggested that these species do not form a monophyletic group. Nonetheless, in that analysis, taxon sampling of nonhybridizing species within Algarobia was relatively scarce. A more complete taxon sampling within Algarobia would give the opportunity to study the timing and evolution of isolating barriers within this group. During recent years, different studies have supported the hypothesis that dry-adapted taxa in different regions of the world diverged concomitantly with the expansion of arid environments. That is the case

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640

PROSOPIS PHYLOGENY AND EVOLUTION for Phylica (Richardson et al., 2001), Ruchoideae (Klak, Reeves & Hedderson, 2003), and Ehrharta (Verboom, Linder & Stock, 2003) in South Africa, Tiquillia (Moore & Jansen, 2006) and Agave (GoodAvila et al., 2006) in North America, and Rheum (Wang, Yang & Liu, 2005) in East Asia. Evidence supporting this hypothesis is in all cases a temporal relationship between the increase of arid environments and the diversication of these groups. Due to the affinity of Prosopis with arid environments, this is a promising hypothesis to be tested in this genus.

623

OBJECTIVES

AND HYPOTHESIS

The present study aimed to investigate the main patterns of Prosopis diversication. We are particularly interested in answering the following questions. (1) Which are the relationships among the main lineages within Prosopis? (2) When did Prosopis diversication occur? (3) Has the spreading of arid environments driven Prosopis evolution, as suggested for other arid-adapted taxa? To answer these questions, a combined approach was applied that included molecular phylogeny, molecular dating, and character optimization analyses.

pair: NIA3F-NIA3R (Howarth & Baum, 2002) and sequenced with a nondegenerate pair: NIAproF (5GAACCAGCAGTTGTTCATCAT-3) and NIAproR (5ACTGGTGCTGGTGTTTTTGG-3). trnL intron was amplied and sequenced using the primers c, d, e and f (Taberlet et al., 1991). Partial sequences of trnK intron were amplied with Ac283R and trnK 3914F primers (Miller & Bayer, 2001). Partial sequences of matK were amplied and sequenced with primers 685F and 1265R (Lavin et al., 2000). To amplify the region trnS-psbC, trnK-matK and trnL-trnF, the following cycling prole was used: 35 cycles of 1 min at 94 C, 1 min at 56 C, and 1.5 min at 72 C. In the case of G3pdh and specic NIA amplication, the cycling prole was the same, except that annealing temperature for these regions was set to 52 C. For the amplication of NIA region with degenerate primers, a stepdown programme was used that included ten cycles in which melting temperature decreased 1 C per cycle starting from 55 C. Twenty additional cycles were performed with 30 s at 94 C, 30 s at 45 C, and 40 s at 72 C. In all cases, the programme started with one cycle at 94 C for 4 min and nished with 7 min at 72 C.

MATERIAL AND METHODS MOLECULAR TECHNIQUES

Five different gene regions were sequenced including two nuclear markers: coding and noncoding sequences of the glyceraldehyde-3-phosphate dehydrogenase gene (G3pdh), one intron of the nitrate reductase gene (NIA) and three chloroplast markers: partial sequences of trnS-psbC, trnK-matK and trnLtrnF regions. Total DNA was extracted from ve day old cotyledons or silica dried leaves and/or stems with the DNeasy Plant Mini Kit (QIAGEN). All polymerase chain reaction reactions were conducted in a total volume of 50 mL, containing 80 ng of DNA template, 1 Taq-Buffer, 2.5 mM MgCl2, 0.25 U Taq polymerase, 0.2 mM of each DNTP and 0.04 mM of each primer. The intergen between the psbC and trnS genes was amplied using the primers psbC and trnS (Demesure, Sodzi & Petit, 1995) together with a new internal primer psbCint (5-CGTTCTTGCCAAGGCTGTAT-3). Since a preliminary survey indicated that most of the variation among sequences was located in the rst 500 bp from the 3 end region of trnS gene, only this region was subsequently sequenced for the rest of the species. G3pdh region was amplied with primers GPDX7F and GPDXR9 (Strand, Leebens-Mack & Milligan, 1997) and a newly designed primer G3pdhintF (5-GACTGGAGGGGTGGAAGAG-3). NIA intron was amplied with a degenerate primer

PHYLOGENETIC

ANALYSIS

Thirty species of Prosopis were included in this analysis. Taxon sampling within Prosopis was planned to include representatives from the entire geographical range of the genus (Appendix 1). Moreover, taxonomic diversity of the genus was totally represented as species of all Sections and Series were included. Three different analyses that involved different combinations of taxa and molecular markers were performed. A rst analysis was performed comprising trnL-trnF and trnK-matK (two-marker analysis) sequences of 80 legume species downloaded from GenBank (see Supplementary material) plus ve Prosopis species that represented each Section of the genus. In the second analysis (three-marker analysis), G3pdh, NIA, and trnS-psbC sequences from all the sampled species of Prosopis plus Prosopidastrum angusticarpum, Acacia caven, Xerocladia viridiramis, and Lotus japonicus were analysed in a total evidence context (Kluge, 1989). As more than one sequence of each species was sampled in Algarobia clade, the total number of terminals was 41. This was due to the high level of shared polymorphism found in previous analyses within this group (Saidman & Vilardi, 1987; Bessega et al., 2000). A third analysis (ve-marker analysis) involved a combination of the rst two datasets. Sequences were edited with Bioedit software (Hall, 1999) and aligned with Clustal X (Thompson et al., 1997) with a posteriori minor manual changes. Par-

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640

624

S. A. CATALANO ET AL. species of Section Algarobia (Prosopis pallida) and one species of genus Prosopidastrum (Prosopidastrum mexicanum) was used to x the node subtending Prosopidastrum and the American species of Prosopis (ASP). To take into account the error associated with the estimation of Lavin et al. (2005), we repeated the analysis considering the mean SD. A second calibration point was used to give minimum age to the ASP divergence. In this case, this fossil corresponds to pollen grains obtained from Early Oligocene sediments of British Columbia, Canada (Piel, 1971). As this Epoch extends from 28.4 and 33.7 Mya, the former age was used to dene the minimum age. Since the sister group of ASP + Xerocladia clade was not clearly dened in our phylogenetic analysis, two different topologies were considered, one of them corresponding to the results of the three-marker analysis and the other corresponding to the two-marker analysis. Error in age estimates was evaluated by bootstrapping following Sanderson (2002).

simony based searches were performed using the TNT program (Goloboff, Farris & Nixon, 2003). Characters and state transformations were considered as equally weighted. Gaps were alternatively treated as a fth state, missing entries or as separate characters Simmons & Ochoterena (2000). Since the results were very similar, only those based on fth state coding are shown and discussed. Within trnL-trnF region, a deletion of approximately 300 bp was found in some Mimosoid species. Individual gaps representing this deletion were treated as missing data. As a modelbased analysis alternative to the parsimony analysis, we also conducted a Bayesian analysis. Search strategies in parsimony and Bayesian analysis are indicated in Appendix 2.

ESTIMATING

DIVERGENCE TIMES

To estimate divergence times, G3pdh and trnS sequences were concatenated and analysed together. Simultaneous analysis of gene sequences from multiple loci was performed because the penalized likelihood method used for age estimation (see below) is prone to errors when dealing with zero length branches or very short branches (Sanderson, 2003). The congruence between both datasets in branch lengths was evaluated by the likelihood ratio test described by Lewis (1998) and Xiang et al. (2005). The statistics of this test is -2 [ln L - (ln L1 + ln L2)], where L1 is the likelihood of the tree obtained from the rst gene, L2 is the likelihood of the tree obtained from the second gene and L is the likelihood obtained from the combined analysis of both genes. The value of this statistic was compared to a c2 distribution with degrees of freedom equal to the sum of 2N + 3 - P, where N is the number of terminals and P is the number of free parameters of the model used in the likelihood calculation. NIA sequences were not included because it was not possible to obtain the sequence of this region from P. angusticarpum, one of the taxa that denes the unique xed node in the calibration step. Furthermore, X. viridiramis was not included because only a part of G3pdh was possible to be sequenced. Branch lengths were estimated by Maximum Likelihood using PAUP* (Swofford, 2002). The evolutionary model was chosen with the hierarchical likelihood ratio test as implemented in Modeltest, version 3.06 (Posada & Crandall, 1998). Penalized likelihood implemented in the R8s software (Sanderson, 2003) was used for age estimation using the Powell algorithm and dening smoothing parameter values with a cross validation procedure (Sanderson, 2002). Divergence times were derived from molecular data combining two calibration points. The time obtained in Lavin et al. (2005) for the divergence between one

DIVERSIFICATION

ANALYSIS

To study the possible changes in diversication rate over time a lineages through time (LTT) analysis (Nee et al., 1995) was performed. LTT analysis was repeatedly performed with extreme values of divergence times obtained in the penalized likelihood analysis. The lack of resolution in the mesquite clade (see Results) was considered in two ways for this analysis: (1) as a hard polytomy with all the lineages diverging simultaneously and (2) as if the diversication occurred at a constant rate. Consequently, PHYLOGEN program, version 1.1 (Rambaut, 2002), was used to simulate trees under a Yule Model of diversication (Yule, 1924) with the nal number of species equal to the present in the mesquite group. Subsequently, this subtree was grafted in the original tree, replacing the polytomy in the mesquite clade. This was repeated for 100 different simulated subtrees. As the general pattern obtained was the same, LTT of one randomly chosen tree is shown. To examine the effects of missing species in LTT analysis, these were placed in the most likely position on the tree on the grounds of taxonomic treatment (Burkart, 1976) in accordnace with Barraclough & Vogler (2002). To compare divergence rates within Prosopis with those obtained in other plant genera, per lineage net diversication rate (NDR) sensu Coyne & Orr (2004) was calculated. As in LTT analysis, missing species were included in their most likely place on the tree.

ESTIMATION

OF ANCESTRAL HABITAT

To study the historical relationship of Prosopis with arid environments, maximum and minimum value of humidity index (HI) were gathered for the natural

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640

PROSOPIS PHYLOGENY AND EVOLUTION

Table 1. Statistics of the different phylogenetic analyses based on parsimony Two-marker analysis Number of MPT Length Number of characters Number of informative characters CI/RI 224 5079 4031 (14442587) 1357 (609748) 0.496/0.655 Three-marker analysis 9 1552 1764 (648701415)* 391 (19318315)* 0.695/0.846

625

Five-marker analysis 4680 6710 5795 1458 0.535/0.700

*Statistics for G3pdh, NIA and trnS-psbC characters, respectively. Statistics for trnL-trnF and trnK-matK characters, respectively. CI, consistency index (excluding uninformative characters); MPT, most parsimonious trees; RI, retention index.

geographical range of the species of Prosopis included in the phylogenetic analysis plus X. viridiramis. HI is dened as the ratio of precipitation to potential evapotranspiration (UNEP, 1992). Species distributions were dened from the bibliography (Ross, 1975; Burkart, 1976; Rzedowski, 1988; Roig, 1993) and from herbarium specimens (BAFC, SI, LIL, CTES). Climate data for each species (see Supplementary material) were extracted from FAO World Climate Database using the New LocClim software (FAO, 2005). Regions were classied as a function of HI values following UNEP (1992): 0.05 < Hyper-Arid; 0.050.20 Arid; 0.200.50 Semi-Arid; 0.500.65 Dry-Subhumid; > 0.65 Humid. The phrase arid environments is used here to collectively represent hyperarid, arid and semiarid classes. Maximum and Minimum of HI (MaHI and MiHI, respectively) were independently optimized on the tree as in the MinMax coding (Hardy & Linder, 2005). Once these values were optimized, they were used to delimit the total range of HI for each ancestral node. We considered that MinMax coding is the most suitable for this character because maximum drought tolerance might be partial or totally decoupled from maximum humidity tolerance. MaHI and MiHI were optimized using the implementation in TNT (Goloboff et al., 2003) of Farris (1970) optimization to deal with continuous characters (Goloboff, Mattoni & Quinteros, 2006). In this case, linear changes are minimized along each branch. To evaluate how uncertainty in the state of the root can affect the results, the analysis was repeated with two extreme values: hyperarid and humid. In addition, the effect of topological uncertainty within the mesquite clade (see Results) in the results was evaluated by considering 10 000 different resolutions of this clade. A strict association of a particular node with arid environments was determined when the value of MaHI was lower than the limit between humid and arid environments (HI < 0.50). A nonstrict association with arid environments corresponded to a value of MiHI lower and MaHI higher than this limit. No association with arid

environment occurred when MiHI value was higher than this limit.

RESULTS
PHYLOGENETIC
RESULTS

Two-marker analysis Statistics of the different phylogenetic analyses based on parsimony are shown in Table 1. Strict consensus suggests that Prosopis is not monophyletic (Fig. 2). Forcing the monophyly of the genus gave a tree nine steps longer than the optimum (Table 2). ASP formed a highly supported clade (BS = 4; J = 95) with X. viridiramis (subsequently named ASP + Xerocladia). ASP did not appear as monophyletic because X. viridiramis formed a monophyletic group with P. tamarugo. However, this clade is not well supported and (BS = 1; J = 66) and forcing the monophyly of ASP gave a tree only one step longer than the optimum. Prosopis cineraria appeared as sister group of this clade though with low support (BS = 1; J < 50). Prosopis nigra of Section Algarobia formed a strongly supported group (BS = 5; J = 98) with Prosopis argentina of Section Monilicarpa. Bayesian analysis gave identical results concerning the lack of monophyly of Prosopis and the relationships among APS. By contrast, P. cineraria does not appear forming a monophyletic clade with ASP + Xerocladia. Posterior probabilities for the groups found in the Bayesian analysis are given in Figure 2. Three-marker and ve-marker analyses Strict consensus obtained in this analysis agreed with two-marker analysis in rejecting Prosopis monophyly (Fig. 3). Forcing the monophyly of the genus gave a tree six steps longer than the optimum (Table 2). As in two-marker analysis, ASP formed a monophyletic group with X. viridiramis (APS + Xerocladia clade), though with lower support (BS = 1; J = 64). In addition, forcing the monophyly of the ASP gave a tree only one steps longer than the optimum. Two

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640

626

S. A. CATALANO ET AL.
Lotus japonicus Cercidium andicola Dinizia excelsa Piptadeniastrum africanum 2/66 Amblygonocarpus andongensis >12/100 0.68 Adenanthera pavonina 1.00 >12/100 >12/99 1/52 Tetrapleura tetraptera 1.00 0.93 Pseudoprosopis gilletii 12/100 Calpocalyx dinklagei 1.00 6/98 3/1.00 Xylia africana Pentaclethra eetveldtiana 1/>12/100 Elephantorrhiza elephantina 1.00 Entada abyssinica 2/Plathymenia reticulata Prosopis africana 9/98 1/55 Fillaeopsis discophora 1.00 Newtonia buchananii 0.78 9/99 Cylicodiscus gabunensis Prosopis cineraria 1/Prosopis argentina 5/98 1/Prosopis nigra 4/95 1.00 0.96 1.00 1/66 Prosopis tamarugo 1.00 Xerocladia viridiramis Neptunia gracilis 1/3/76 Prosopidastrum angusticarpum 0.99 4/98 1.00 1.00 Prosopidastrum mexicanum 1/Calliandropsis nervosus 8/100 Gagnebina bernieriana 4/87 1.00 4/91 Alantsilodendron alluaudianum 1/1.00 Dichrostachys paucifoliolata 1/74 Mimozyganthus carinatus 1.00 Piptadeniopsis lomentifera 1/2/55 Leucaena greggii 0.98 1/90 Kanaloa kahoolawensis 0.99 5/100 Desmanthus acuminatus 1.00 5/95 Schleinitzia insularum Anadenanthera colubrina Acacia schottii Acacia caven >12/100 Acacia constricta 1.00 Acacia neovernicosa 11/100 Acacia karroo 3/73 1.00 Acacia tortilis Acacia hindsii 8/100 2/78 Acacia chiapensis 1/72 1.00 2/78 0.99 Acacia collinsii 1.00 0.53 Acacia cochlicantha 3/93 Acacia macracantha 1.00 Acacia pennatula 1/58 Parkia biglandulosa Piptadenia viridiflora Mimosa aculeaticarpa 2/79 Microlobius foetidus 6/99 1/Parapiptadenia rigida 1.00 Acacia modesta >12/100 1.00 Acacia senegal 1/4/92 Acacia berlandieri 1.00 6/74 0.79 Acacia roemeriana 1/1.00 3/75 Acacia glomerosa 1.00 Acacia schweinfurthii Acacia dolichostachya 6/97 8/Acacia coulteri 1.00 1.00 Acacia visco Albizia kalkora Cathormion umbellatum Enterolobium contortisiliquum Inga edulis 1/Paraserianthes lophantha 1.00 2/Acacia spinescens Chloroleucon mangens 10/100 Calliandra carbonaria 1.00 Cedrelinga cateniformis 9/100 Ebenopsis ebano 1/1.00 Havardia albicans 1.00 1/Acacia ampliceps 5/75 Acacia drummondii 1.00 Acacia pulchella Acacia boliviana 1/1/Faidherbia albida 0.68 Zapoteca tetragona Acacia melanoxylon 7/92 1/68 Acacia elata 1.00 0.98 Acacia platycarpa Pararchidendron pruinosum 1/3/Samanea saman 4/Lysiloma divaricatum 0.55 1.00 Pseudosamanea guachapele

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640

PROSOPIS PHYLOGENY AND EVOLUTION

627

Figure 2. Strict consensus of the 224 optimal trees based two-marker analysis (trnL-trnF, trnK-matK) with parsimony as optimality criteria. Absolute Bremer support (left) and jackkning values over 50% (right) are indicated above branches. Posterior probabilities estimated in Bayesian analysis are indicated below branches. Prosopis species are underlined.

Table 2. Extra steps obtained in searches conducted with positive constrains for some groupings not appearing in most parsimonious trees Two-marker analysis (5079) Section Strombocarpa + Prosopis cineraria Prosopis + Prosopidastrum Prosopis + Xerocladia Genus Prosopis Section Algarobia OSP + Xerocladia OSP ASP 6 14 4 9 NA 8 4 1 Three-marker analysis (1552) 63 3 5 6 10 6 5 1 Five-marker analysis (6710) 68 14 5 11 10 11 11 3

In parentheses: optimal lengths in unconstrained searches. ASP, Americans species of Prosopis; NA, not applicable; OSP, Old World species of Prosopis.

subclades were recognized in ASP + Xerocladia clade. One of them is formed by species of Section Strombocarpa plus X. viridiramis whereas the other (subsequently named Algarobia s.l.) is formed by species of Section Algarobia plus P. argentina of Section Monilicarpa (BS = 6; J = 100). Section Algarobia appeared as paraphyletic because P. argentina formed a well supported clade (BS = 5; J = 100) with species of Series Humiles, Sericanthae, and Denudantes of Section Algarobia (Prosopis kuntzeiP. argentina clade). Forcing the monophyly of Section Algarobia gave a tree ten steps longer. Series Denudantes (represented here by Prosopis denudans, Prosopis ruizleali and Prosopis castellanosii) appeared as monophyletic but Series Sericanthae (P. kuntzei and Prosopis sericantha) would be paraphyletic. The sister group of the P. kuntzeiP. argentina clade was a very well supported clade (BS = 6, J = 100) formed by species of series Chilenses, Pallidae, and Ruscifoliae (mesquite clade). However, the relationships among the species of this clade are unclear because the strict consensus is highly unresolved and individuals of the same species in most of the cases did not form monophyletic groups. Section Strombocarpa was monophyletic with X. viridiramis as its sister clade. Within this section, two groups were well supported. One of them corresponded to Series Cavenicarpae (Prosopis ferox and P. tamarugo) whereas the other was formed by two North American species of Series Strombocarpae: Prosopis pubescens and Prosopis palmeri. Bayesian analysis showed very similar results (Fig. 3). The

main difference was related with the sister group of APS + Xerocladia. In this analysis, the sister group was formed by A. caven and P. angusticarpum. The results obtained in the ve-marker analysis were very concordant with those obtained in the other two analyses (see Supplementary material) and the support increased in several relevant nodes: ASP + Xerocladia clade (BS = 9; J = 96); Strombocarpa clade (BS = 2; J = 83); Strombocarpa + X. viridiramis clade (BS = 2; J = 77). However, differing from the two-marker analysis, P. cineraria did not appear as sister group of APS + Xerocladia. Strict consensus indicated identical relationships within APS + Xerocladia clade than those given by the threemarker analysis, except that the groupings within the mesquite clade are less resolved.

MOLECULAR

DATING

The optimal model chosen by the hierarchical likelihood ratio test was HasegawaKishinoYano plus Gamma (HKY + G). A likelihood ratio test signicantly rejected the hypothesis of evolutionary rate constancy for the molecular regions analysed (Clock constrained lnL = 4394.64; unconstrained lnL = 4363.50; d.f. = 39; P < 0.05). Therefore, divergence times were estimated in a penalized likelihood approach that does not assume rate constancy. To study the possible effect of short branches within mesquite clade in the age estimated for the other nodes outside this group, an analysis leaving only one

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640

628

S. A. CATALANO ET AL.
Lotus japonicus Acacia caven P. cineraria
3 / 60

P. africana
3/ -

1 2

Prosopidastrum angusticarpum Xerocladia viridiramis

4 / 100 0.99 1 / 54 0.64 1 / 64 0.55

2/ -

2 /76 0.91

3
5 /100 0.96 2 / 70 0.58 1 /57 0.63

P. strombulifera P. palmeri P. pubescens 2 / 76 P. ferox 0.99 P. tamarugo P. reptans 3 / 57 0.83 P. torquata P. denudans P. castellanosii P. ruizleali P. argentina 5 P. sericantha 2 / 72 P. humilis 0.69 6 /100 P. kuntzei 1 1.00 4 / 99 1.00 P. kuntzei 2 P. campestris 1 P. campestris 2 P. chilensis 1 P. chilensis 2 P. tamaulipana P. vinalillo 2 P. affinis P. alba P. glandulosa P. juliflora 2 P. alpataco P. ruscifolia 2 / 65 P. caldenia 2 0.96 P. nigra 2 P. caldenia 1 P. flexuosa 2 P. juliflora 1 P. nigra 1 P. vinalillo 1 1 / 62 P. articulata P. flexuosa 1
5/ 96 1.00

P. kuntzei P. argentina clade

Strombocarpa clade

Strombocarpa + Xerocladia clade

1/0.58

6 / 100 0.85

1 / 61

1/ 0.61

Figure 3. Strict consensus of the nine optimal trees obtained in the three-marker analysis (trnS-psbC, G3pdh, NIA) with parsimony as optimality criteria. Absolute Bremer support (left) and jackkning values over 50% (right) are indicated above branches. Posterior probabilities estimated in Bayesian analysis are indicated below branches. Rectangles delimit each Section: 1, Prosopis; 2, Anonychium; 3, Strombocarpa; 4, Algarobia; 5, Monilicarpa. Black bars indicate the clades whose names are used throughout the manuscript.

terminal was performed. The results differ only slightly (data not shown). The common and most relevant pattern observed in the different analyses was (Fig. 4; Table 3): (1) an ancient divergence between the two main lineages of ASP within the Oligocene; (2) a more recent divergence, starting in the late Miocene, within Strombocarpa and Algarobia s.l.; and (3) a divergence among mesquite species starting in the Pliocene.

DIVERSIFICATION

ANALYSIS

Phylogenetic analyses indicated that Old World species of Prosopis are not closely related with the ASP. This was true for P. africana in all the analyses and for P. cineraria in two out of three analyses.

Taking this into account, the study of diversication and climate affinity evolution was restricted to the ASP + Xerocladia clade. Lineage through time analysis clearly suggests an increase in the diversication rate of the ASP group from the Late Miocene (Fig. 5A, B) until the present. Within this period, a more detailed examination suggests two different moments of rate acceleration: one in the Late Miocene and the other in the Pliocene. This general pattern is neither affected by the inclusion of missing species in the analysis (Fig. 5B), nor by the different topologies used to derive the chronograms (data not shown). The NDR estimated for different clades within Prosopis and for other taxa is shown in Table 4.

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640

Mesquite clade

3 / 80 0.99 6 /100 3 / 87 0.90 1.00

American species of Prosopis + Xerocladia clade

Algarobia s.l clade

PROSOPIS PHYLOGENY AND EVOLUTION

Prosopidastrum angusticarpum P. vinalillo 1 P. nigra 1 P. chilensis 1 P. campestris 1 P. chilensis 2 P. affinis P. alba P. alpataco P. ruscifolia P. caldenia 2 P. nigra 2 P. campestris 2 P. vinalillo 2 P. tamaulipana P. glandulosa P. juliflora 2 P. juliflora 1 P. caldenia 1 P. articulata P. flexuosa 1 P. flexuosa 2 P. argentina P. sericantha P. humilis P. kuntzei 1 P. kuntzei 2 P. denudans P. castellanosii P. ruizleali P. strombulifera P. ferox P. tamarugo P. reptans P. torquata P. palmeri P. pubescens

629

OLIGOCENE
Early Late Early

MIOCENE
Middle Late

PLIOCENE PLE

33.7

28.4

23.8

16.4

11.2

5.3

1.8

Figure 4. Chronogram resulting from penalized likelihood analysis. The topology corresponds with one of the most parsimonious trees obtained in three-marker analysis (Fig. 3) and calibrated according to divergence times derived from Lavin et al. (2005). The approximate temporal spreading of arid zones in America is shown over the timescale with darker zones representing the moment of its maximum historical extension. Grey bars over the nodes represent 95% condence interval for node ages obtained by bootstrap techniques. PLE, Pleistocene. Table 3. Divergence times (Mya) obtained in the penalized likelihood analysis Outgroup placement MRCA ASP + Xerocladia Algarobia s.l. Mesquites Strombocarpa Three-marker analysis 28.96 7.89 3.65 9.21 (26.2531.68) (7.158.62) (3.313.99) (8.3510.07) Two-marker analysis 29.37 7.96 3.69 9.31 (27.3933.05) (7.589.14) (3.384.08) (9.1811.07)

Ages estimated xing the node subtending ASP + Xerocladia clade and Prosopidastrum to 33.2 Mya following the mean value obtained in Lavin et al. (2005). Ages estimated considering the mean one standard deviation are given in parentheses. When the topology corresponded to the two-marker analysis the node subtending ASP + Xerocladia node was constrained to a minimum age of 28.4 Mya following the fossil evidence. ASP, American species of Prosopis; MRCA, most recent common ancestor.

ANCESTRAL

CLIMATE ESTIMATION

Optimization analysis unambiguously indicated that the range of the most recent common ancestor (MRCA) of the clade formed by the ASP plus X. viridiramis extended to semiarid or more xeric conditions

(Fig. 6). A strict association of this node with arid environments was obtained for all the reconstructions when the state of the root was dened as hyperarid and, in some of the reconstructions, when the state of the root was dened as humid. In the case of Strombocarpa clade, the different assignments of states to

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640

630
A

S. A. CATALANO ET AL. tionship among Prosopis species and X. viridiramis has been previously indicated by Luckow et al. (2004). However, considering that only species of Section Algarobia were included in that analysis, a closer relationship of ASP with X. viridiramis than with Old World Prosopis species is a novel and interesting result. Prosopis monophyly was also rejected due to the position of P. africana and, in some of the analyses, the position of P. cineraria. It is suggestive that many of the characters used to dene the genus by Burkart (1976) would be, as stated by him, primitive within the whole subfamily (symplesiomorphic). If this is conrmed, the idea of Prosopis as natural group might also be questioned at morphological grounds. A closer relationship among species of the American sections of Prosopis (Strombocarpa, Algarobia, and Monilicarpa) than with species of the Old World sections agrees with the subgeneric classication proposed by (Guinet & Bessedik, 1984) based mainly on palynological evidence. Particularly, they included all the ASP in the subgenus Neoprosopis, whereas the other two subgenera corresponded to Section Prosopis and Anonychium, the two Old World sections considered in Burkart (1976). However, the correspondence between the groups obtained in our analyses and the Subgenus Neoprosopis is not strict because ASP formed a monophyletic group with X. viridiramis. This close relationship among American Sections of Prosopis contrasts with the idea stated by Burkart (1976) that Section Strombocarpa is closely related to Section Prosopis because species of both Sections present comparatively smaller legumes and share the ability to spread by means of rootsuckers. Species of Section Algarobia and P. argentina formed a highly supported clade (Algarobia s.l.) that is subsequently divided into two highly supported clades. One of them corresponded to species of Series Pallidae, Chilenses, and Ruscifoliae (mesquite clade), whereas the other corresponded to species of series Sericanthae, Humiles, and Denudantes plus P. argentina of Section Monilicarpa (P. kuntzeiP. argentina clade). This latter clade, endemic to southern South America, includes all the subaphyllous and aphyllous species sampled for the phylogenetic analysis. This result suggests that this condition might have originated only once in Prosopis history. The results obtained in the present study are consistent with the only previous cladistic analysis involving Prosopis species (Bessega et al., 2006) in showing that the Section Algarobia is not monophyletic, and that P. argentina (Section Monilicarpa) is close to species of Section Algarobia. However, several disagreements between these two studies can be observed. Within Section Algarobia, the clade formed

3.5 3.0 2.5 2.0 1.5 1.0

3.0 2.5 2.0 1.5 1.0

OLIGOCENE
E L E

MIOCENE
M L

0.5
PLIOC PLE

33.7

28.4

23.8

16.4

11.2

5.3

1.8

Figure 5. Lineages-through-time plot (LTT) derived from the chronogram of Fig. 4. A, LTT including missing species. B, LTT with species included in the three-marker analysis. Dotted lines represent LTT when the polytomy in the mesquite clade is considered as hard. Black arrowheads indicate the beginning of the diversication within Algarobia s.l. and Strombocarpa clades. Empty arrowheads indicate the start of mesquite diversication. E, Early; M, Medium; L, Late; PLIOC, Pliocene; PLE, Pleistocene. The approximate temporal spreading of arid zones in America is shown over the timescale with darker zones representing the moment of its maximum historical extension.

the root did not affect the results as, in all cases, its MRCA appear to be restricted to arid environments. The same result was obtained for the MRCA of X. viridiramis + Strombocarpa clade. In the case of the MRCA of Algarobia s.l., P. kuntzeiP. argentina and mesquite clades, optimal reconstructions included in some cases ranges that fell completely within arid conditions whereas, in others, the ranges extended from arid to humid conditions.

DISCUSSION SYSTEMATICS
The phylogenetic analysis performed in the present study is the rst to include species of all the Sections and Series of the genus and intends to test Prosopis monophyly by incorporating several species of related genera. The results of this analysis suggest that Prosopis would not be monophyletic. ASP appear to be more related with X. viridiramis, the only species of this southern Africa genus, than with Old World representatives of Prosopis. A close rela-

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640

Ln (number of lineages)

3.5

Ln (number of lineages)

PROSOPIS PHYLOGENY AND EVOLUTION

631

Table 4. Comparison of net diversication rates (NDR) estimated within Prosopis with those obtained for other plant taxa NDR Prosopis Mesquite clade Prosopis kuntzeiProsopis argentina clade Strombocarpa clade ASP + Xerocladia clade Others taxa Inga1 Lupinus2 Ruschioideae3 Clarkia, section Peripetasma4 Gentianella5 Valeriana6 Yucca7 Ehrharta8 Gaertnera9 Hawaiian silverwoods10 Agave7 Life forms

0.580.72* 0.160.21* 0.270.37* 0.120.14* 0.50 1.923.84 0.771.75 0.62 1.73.2 0.81.3 0.210.27 0.120.57 0.710.83 0.55 0.51

Trees Trees and shrubs Trees and shrubs Trees and shrubs Trees Postrate herbs to treelet Stone plants to tree-like shrubs Herbs Herbs and small shrubs Herbs Succulent rossete plants to arborescent Herbs Shrubs and small trees Vines and shrubs to trees Succulent rossete plants to arborescent

*Range of rates values estimated with times derived from the different calibration strategies. 95% of the species are trees or present arboreal forms. 1 Richardson et al. (2001); 2Hughes & Eastwood (2006); 3Klak et al. (2003); 4Hey (1992); 5Von Hagen & Kadereit (2001); 6 Bell & Donoghue (2005); 7Good-Avila et al. (2006); 8Verboom et al. (2003); 9Malcomber (2002); 10Baldwin & Sanderson (1998). ASP, Americans species of Prosopis.

Mesquites

SA-SH-HU AR-SA SA-SH-HU AR-SA SA-SH-HU AR-SA

SA-SH-HU AR-SA SA-SH-HU AR-SA

P. kuntzeiP. argentina

SA-SH-HU AR-SA SA-SH-HU AR-SA

SA HA-AR

Xerocladia

Xerocladia

Strombocarpa

SA HA-AR

AR-SA AR

Figure 6. Reconstruction of ancestral states of relevant nodes for maximum (above) and minimum (below) humidity index (MaHI and MiHI, respectively). Instead of HI values, classes that represent each value are shown. A, optimization result when the state of the root was dened as hyperarid. B, optimization result when the state of the root was dened as humid. The presence of ambiguities in the optimization step and/or the different possible resolutions within the mesquite clade produced, in some nodes, more than one optimal value for MiHI and MaHI.

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640

Strombocarpa

SA HA-AR

SA AR

P. kuntzeiP. argentina

Mesquites

632

S. A. CATALANO ET AL. of Section Strombocarpa. The age of this fossil is compatible with the times estimated in the molecular dating analysis as both indicate that the divergence between the to major groups of Prosopis occurred before the middle Miocene. A more detailed comparison of divergence times will require the combined inclusion of fossil and extant species in a phylogenetic analysis.

by species of Series Ruscifoliae, Chilenses, and Pallidae (mesquite clade) is not monophyletic in the analysis of Bessega et al. (2006). A similar situation is observed in Section Strombocarpa, which is monophyltic in our analysis (Fig. 3) but not in that of Bessega et al. (2006). In addition, in the present analysis, we found that American Prosopis species are more closely related among themselves than with Old World Prosopis species, whereas in the analysis of Bessega et al. (2006), P. cineraria, an Old World species, appears intermingled with New World species. The groupings obtained in our analysis are generally more in agreement with previous analyses (Ramrez et al., 1999; Bessega et al., 2005), and with traditional taxonomy (Fig. 3). Although several ad hoc explanations could be given to account for the disagreement between these two analyses, it is possible that the wider sampling of species and molecularmarkers performed in the present study might be the best explanation for these differences.

MESQUITES

DIVERSIFICATION

MOLECULAR

DATING AND FOSSIL RECORD

Molecular dating analysis suggests that the divergence between the two main groups within ASP + Xerocladia clade is remarkably ancient (26.233 Mya) considering that the diversication of the extant members of the whole subfamily Mimosoideae would have occurred during the last 40 Myr (Lavin et al., 2005). However, most of the process of diversication within each of the major clades occurred more recently, in the Late Miocene. This contrast between early divergence of the main groups and late divergence within these groups had already been established on the grounds of morphological evidence (Burkart & Simpson, 1977; Pasiecznik et al., 2001). The number of fossil remains assigned to Prosopis is particularly abundant (Catalano et al. unpubl. data). However, as none of the fossils have been analysed in a phylogenetic context and some of the descriptions are poorly detailed, their assignation to Prosopis and its subgroups is, in some cases, very tentative. The oldest fossil that could be attributed to the ASP clade (Guinet & Bessedik, 1984), which was used as one of the calibration points in our molecular dating analysis, belongs to pollen grains (Prosopis quesneli) of the early Oligocene from British Columbia, Canada (Piel, 1971). A similar age has been estimated for fruit remains (Prosopis lazarii) found in the palaeoora of Puebla, Mexico (Magallon-Puebla & Ceballos-Ferriz, 1994). Nevertheless, the affinity of this fossil with extant groups within the genus is uncertain. Recently, Anztegui & Herbst (2004) have described, based on leaets remains, a new fossil species of Prosopis from the Middle Miocene of Argentina that appears to be related with extant members

The results obtained in the present study strongly support a close relationship of species of Series Chilenses, Pallidae, Ruscifoliae (mesquite clade). Interestingly, many of these species were supposed to form a syngameon (Palacios & Bravo, 1981) because of their ability to hybridize naturally, their high morphological similarity and partially overlapping geographical distributions. However, several recent analyses suggest that hybrid formation does not produce signicant gene ow among these species. First, molecular studies showed that, in spite of their high genetic similarities, populations formed groupings coincident with their specic origin (Palacios & Bravo, 1981; Saidman & Vilardi, 1987; Burghardt, 1995; Bessega et al., 2000). This was true even for populations of different species that were sympatric. In addition, Bessega et al. (2000) showed that sympatric species are not genetically more similar than allopatric ones. Second, the only analysis of hybrid viability available to date (Naranjo, Poggio & Enus Zeiger, 1984) suggests that, at least for some species combinations, hybrids are partially or totally sterile. In the light of our results, the high genetic similarity among these species might now be explained by their recent divergence (< 4 Mya) without the need to invoke introgression among them. The average per lineage diversication rate estimated for the mesquite group is comparable to other known rapid plant radiations (Table 4). In particular, the diversication rate estimated for this group is found to be higher than in Inga, another genus from the same subfamily that also comprises tree species. It is possible that this rapid and recent radiation of mesquites might be the cause of the lack of resolution obtained in the mesquite clade in the present study. A similar explanation for the lack of resolution has been proposed in diverse plant genera (Hodges & Arnold, 1995; Baldwin & Sanderson, 1998; Richardson et al., 2001; Hughes & Eastwood, 2006).

EVOLUTION

OF REPRODUCTIVE ISOLATING

MECHANISM WITHIN ALGAROBIA

Molecular dating and phylogenetic results obtained in the current study provides an improved insight into

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640

PROSOPIS PHYLOGENY AND EVOLUTION the evolution of isolating mechanism within the genus. In the case of the mesquite clade, as previously indicated, gene ow among its species would be highly restricted. This suggests the existence of isolating mechanism(s) that prevent introgression among them. The existence of hybrid between these species in nature indicates that prezygotic reproductive isolation barriers are probably weak (Palacios & Bravo, 1981) and stresses the potential importance of postzygotic isolating mechanisms within this group. The observation that hybrids between mesquite species are generally found in disturbed environments (Palacios & Bravo, 1981; Verga, 1995) supports the idea that hybrids are not intrinsically inviable, but that their development is limited by the lack of suitable environmental conditions (ecological inviability; Coyne & Orr, 2004). Evidence for a second postzygotic isolating barrier has been obtained by Naranjo et al. (1984) who observed diminished fertility in some hybrid combinations. Given the times derived from molecular dating, these postzygotic isolating mechanisms would have developed in less than 4 Myr, the total span for the diversication of this group (Table 3). The phylogenetic results suggest that the ability to hybridize is not a characteristic extended throughout the Algarobia s.l. group. Indeed, the species of the two main clades found within this group (Fig. 3) present strikingly different behaviour. Whereas, as previously indicated, mesquites do frequently hybridize, hybrid formation between species of the P. kuntzei P. argentina clade has not been documented. Although species of the P. kuntzeiP. argentina clade might have developed isolating mechanisms that prevent hybrid formation, we consider that the reduced level of sympatry between most of these species is a sufficient explanation for the difference. Furthermore, no hybrids have been observed between mesquite species and species of P. kuntzeiP. argentina clade even if in this case they frequently appear in sympatry. This suggests the development of another barrier either intrinsic postzygotic or prezygotic in less than 8 Myr (Table 3), the estimated divergence time between these two clades.

633

analysis performed suggested an ancient association of Prosopis with arid conditions (Fig. 6), already established before the splitting of the two main groups within ASP + Xerocladia clade (26.733 Mya). This result disagrees with Roig (1993) who proposed that extant species of Prosopis occupying arid regions in America have originated from ancestral lineages of this genus which were restricted to wetter regions. Palaeoenvironmental data suggest that, although the development of large arid areas in America started in the late Miocene (see below), there were local arid conditions in North America (Axelrod, 1979a) and South America (Volkheimer, 1971; Jordan, Schlunegger & Cardozo, 2001) during much of the Tertiary. It is possible therefore that ancestral lineages of Prosopis had occupied these sites until the later spreading of arid conditions.

DIVERSIFICATION

PATTERN AND CAUSES

ANCESTRAL

CLIMATE RECONSTRUCTION

During recent years, optimization techniques have been increasingly used to infer ancestral ecological conditions (Verboom et al., 2003; Graham et al., 2004; Hardy & Linder, 2005; Schrire, Lavin & Lewis, 2005). In the present study, an optimization analysis was performed to evaluate the historical association of Prosopis with arid environments. In particular, we were interested in dening since when Prosopis species have occupied this kind of environment. The

One of the predictions derived from the hypothesis of aridity-driven diversication of Prosopis was a temporal correlation between the spreading of arid areas in the Americas and the diversication of the genus. Molecular clock dating showed that, except for the splitting between the two main groups, the divergence among extant lineages within ASP clade would have occurred in the Late Miocene or more recently, coincident with the estimated spreading of arid environments in America. Although with distinctive characteristics in each case, the development of arid environments occur simultaneously in North and South America starting in the Late Miocene and continuing in the Pliocene and Pleistocene. In North America, the development of regional arid areas is thought to have started in the late Miocene to Pliocene (Graham, 1999; Riddle & Hafner, 2006) associated with the uplift of the plateaus in Western North America. The Sonora desert, where some species of the genus are present today, would have developed from 15 to 8 Mya (van Devender, 2000). This process of aridication continued until the Pleistocene, reaching a considerable distribution in the interglacials (Axelrod, 1979a,b). In South America, the spreading of large areas of semiarid-arid conditions possibly started in the late Miocene concomitantly with the uplift of the Andes in the Quechua distrophic phase when it reached half of its present elevation (Gregory-Wodzicki, 2000) and started to act effectively as a barrier to moisture laden winds. During the Diaguita diastrophic Phase, the nal uplift of the Pampean Mountain Range and the Central Andes produced a rain-shadow effect that resulted in the extremely xeric conditions existing at present on the areas located between them (Pascual, Ortiz-Jaureguizar & Prado, 1996; Alberdi. Bonadonna

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640

634

S. A. CATALANO ET AL. grants to B.O.S. from University of Buenos Aires (UBACYT_X321) Agencia Nacional de Promocin Cientca y Tecnolgica (PICT 32064/05), and a grant to J.C.V. from CONICET (PIP 5122/04) grants. The authors wish to dedicate this manuscript to the memory of our mentor and advisor Professor Carlos A. Naranjo. Santiago A. Catalano wishes also to dedicate this manuscript to the memory of his father.

& Ortiz-Jaureguizar, 1997). These climatic changes would have produced the establishment of the biomes where Prosopis is established today (Axelrod, Kalin Arroyo & Raven, 1991; Pascual et al., 1996). The possible correlation of the ASP diversication with the expansion of arid environments is also supported by the analysis of the temporal changes of diversication rates. LTT analysis indicates a clear increase in diversication rates from the late Miocene until the present compared to previous times. When this pattern is analysed in detail, two different shifts towards higher rates may be distinguished. The rst of those occurred in the late Miocene and is coincident with the start of the divergence within each of the two major lineages of the ASP. The second shift appears to be associated with the start of mesquites divergence in the Pliocene and is in agreement with the high diversication rate estimated for this group (Table 4). A wrong assignment of missing taxa could have potentially affected the conclusion derived from this analysis. However, as most of the species not sampled belong to Series Chilensis, Ruscifoliae, and Pallidae, which formed the highly supported mesquite clade, a wrong assignment seems to be improbable. The combined assessment of the results obtained in the present study is in agreement with the hypothesis that Prosopis evolved and diversied concomitantly with the spreading of arid areas in the Americas. This result is coincident with those obtained for other dry-lover plant groups distributed almost exclusively in North America, which is suggested to have undergone a parallel radiation in the last 10 Myr (GoodAvila et al., 2006). To our knowledge, no previous study found this pattern for a group with mainly South American distribution, as is the case of Prosopis. Consequently, the results of the present study suggest an extension of this parallel radiation to a continental scale.

REFERENCES
Alberdi MT, Bonadonna FP, Ortiz-Jaureguizar E. 1997. Chronological correlation, paleoecology, and paleobiogeography of the Late Cenozoic South American Rionegran landmammal fauna: a review. Revista Espaola de Paleontologa 12: 249255. Anztegui LM, Herbst R. 2004. Megaora (hojas y frutos) de la Formacin San Jos (Mioceno Medio) en ro Seco, departamento Santa Mara, provincia de Catamarca, Argentina. Ameghiniana 41: 423436. Axelrod DI. 1979a. Age and origin of Sonoran desert vegetation. Occasional Papers of the California Academy of Sciences 132: 174. Axelrod DI. 1979b. Desert vegetation, its age and origin. In: Goodin JR, Northington DK, ed. Arid land plant resources. Lubbock, TX: International Center for Arid and Semiarid Land Studies, 172. Axelrod D, Kalin Arroyo M, Raven P. 1991. Historical development of temperate vegetation in the Americas. Revista Chilena de Historia Natural 64: 413446. Baldwin BG, Sanderson MJ. 1998. Age and rate of diversication of the Hawaiian silversword alliance (Compositae). Proceedings of the National Academy of Sciences of the United States of America 95: 94029406. Barraclough TG, Vogler AP. 2002. Recent diversication rates in North American tiger beetles estimated from a dated mtDNA phylogenetic tree. Molecular Biology and Evolution 19: 17061716. Bell CD, Donoghue MJ. 2005. Phylogeny and biogeography of Valerianaceae (Dipsacales) with special reference to the South American valerians. Organisms, Diversity and Evolution 5: 147159. Bessega C, Ferreyra LI, Vilardi JC, Saidman BO. 2000. Unexpected low genetic differenatiation among allopatric species of Section Algarobia of Prosopis (Leguminoseae). Genetica 109: 255266. Bessega C, Saidman BO, Vilardi JC. 2005. Genetic relationships among American species of Prosopis (Leguminosae) based on enzyme markers. Genetics and Molecular Biology 28: 277286. Bessega C, Vilardi JC, Saidman BO. 2006. Genetic relationships among American species of the genus Prosopis (Mimosoideae, Leguminosae) inferred from ITS sequences: evidence for long-distance dispersal. Journal of Biogeography 33: 19051915. Bremer K. 1994. Branch support and tree stability. Cladistics 10: 295304.

ACKNOWLEDGEMENTS
We thank P. Steibel, H. Troiani, A. Vilela, and G. Humphreys for helping in sampling; N. Iusem and L. Maskin for helping in laboratory issues; F. Prevosti for helpful comments on the manuscript; E. Ulibarri for helping in SI herbarium; P. Harris for P. africana material; O. Ruiz for L. japonicus material; J. Juarez Muoz for P. palmeri, Prosopis tamaulipana and Prosopis articulata material; C. Pometti for A. caven material; M. Luckow for lending P. nigra and X. viridiramis sequences and X. viridiramis DNA; A. Gottlieb for helping in different steps of this study. S.A.C. is fellow of Consejo Nacional de Investigaciones Cientcas y Tcnicas (CONICET, Argentina). B.O.S. and J.C.V. are members of Carrera del Investigador Cientco of CONICET. This study was supported by

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640

PROSOPIS PHYLOGENY AND EVOLUTION

Burghardt AD. 1995. La identidad de cuatro especies de Prosopis L. expresada a travz de sus patrones electroforeticos. Mendeliana 12: 3850. Burkart A. 1976. A monograph of the genus Prosopis (Leguminosae subFam Mimosoideae). Journal of the Arnold Arboretum 57: 219249. Burkart A, Simpson BB. 1977. The genus Prosopis and annoted key to the species of the world. In: Simpson B, ed. Mesquite: its biology in two desert ecosystems. Stroudsburg, PA: Dowden, Hutchinson & Ross, 201215. Coyne JA, Orr HA. 2004. Speciation. Sunderland, MA: Sinauer Associates. Demesure B, Sodzi N, Petit RJ. 1995. A set of universal primers for amplication of polymorphic non-coding regions of mitochondrial and chloroplast DNA in plants. Molecular Ecology 4: 129131. van Devender TR. 2000. The deep history of the Sonoran Desert. In: Phillips SJ, Comus PW, eds. A natural history of the Sonoran desert. Tucson, AZ: Arizona-Sonora Desert Museum Press; Berkeley, CA: University of California Press, 6169. FAO. 2005. New LocClim 1.06 local climate estimator. Agrometeorology group. Rome: FAO/SDRN. Farris S. 1970. Methods for computing Wagner trees. Systematic Zoology 34: 2124. Goloboff PA. 1999. Analyzing large data sets in reasonable times: solutions for composite optima. Cladistics 15: 415 428. Goloboff PA, Farris JS, Nixon K. 2003. TNT: tree analysis using new technology. Software and documentation available at http://www.zmuc.dk/public/phylogeny Goloboff PA, Mattoni CI, Quinteros AS. 2006. Continuous characters analyzed as such. Cladistics 22: 589601. Good-Avila SV, Souza V, Gaut BS, Eguiarte LE. 2006. Timing and rate of speciation in Agave (Agavaceae). Proceedings of the National Academy of Sciences of the United States of America 103: 91249129. Graham A. 1999. Late cretaceous and cenozoic history of North American vegetation. New York, NY: Oxford University Press. Graham CH, Ron SR, Santos JC, Schneider CJ, Moritz C. 2004. Integrating phylogenetics and environmental niche models to explore speciation mechanisms in Dendrobatid frogs. Evolution 58: 17811793. Gregory-Wodzicki KM. 2000. Uplift history of the central and northern Andes: a review. Geological Society of America Bulletin 112: 10911105. Guinet PH, Bessedik M. 1984. Presence de Genre Prosopis (Leguminosae-Mimosoideae) a lAquitanien Basal dans laude (Languedoc-France). Pollen et Spores 26: 101108. Hall TA. 1999. BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/ NT. Nucleic Acids Symposium Series 41: 9598. Hardy CR, Linder HP. 2005. Intraspecic variability and timing in ancestral ecology reconstruction: a test case from the Cape ora. Systematic Biology 54: 299316. Herendeen PS, Crepet WL, Dilcher DL. 1992. The fossil history of the leguminosae: phylogenetic and biogeographic

635

implications. In: Hendereen PS, Dilcher DL, eds. Advance in legume systematics. Kew: The Royal Botanical Garden, 313 316. Hey J. 1992. Using phylogenetic trees to study speciation and extinction. Evolution 46: 627640. Hodges SA, Arnold ML. 1995. Spurring plant diversication: are oral nectar spurs a key innovation? Proceedings of the Royal Society of London Series B, Biological Sciences 262: 343348. Howarth DG, Baum DA. 2002. Phylogenetic utility of a nuclear intron from nitrate reductase for the study of closely related plant species. Molecular Phylogenetics and Evolution 23: 525528. Hughes CE, Eastwood R. 2006. Island radiation on a continental scale: exceptional rates of plant diversication after uplift of the Andes. Proceedings of the National Academy of Sciences of the United States of America 103: 1033410339. Jordan TE, Schlunegger F, Cardozo N. 2001. Unsteady and spatially variable evolution of the Neogene Andean Bermejo foreland basin, Argentina. Journal of South American Earth Sciences 14: 775798. Klak C, Reeves G, Hedderson T. 2003. Unmatched tempo of evolution in Southern African semi-desert ice plants. Nature 427: 6365. Kluge AG. 1989. A concern for evidence and a phylogenetic hypothesis of relationships among Epicrates (Boidae, Serpentes). Systematic Zoology 38: 725. Lavin M, Thulin M, Labat J, Pennington RT. 2000. Africa, the odd man out: molecular biogeography of Dalbergioid legumes (Fabaceae) suggests otherwise. Systematic Botany 25: 449467. Lavin M, Herendeen PS, Wojciechowski MF. 2005. Evolutionary rates analysis of Leguminosae implicates a rapid diversication of lineages during the Tertiary. Systematic Biology 54: 575594. Lewis PO. 1998. Maximum likelihood as an alternative to parsimony for inferring phylogeny using nucleotide sequence data. In: Sotis DE, Soltis PS, Doyle JJ, eds. Molecular systematics of plants II DNA sequencing lluwer. Boston, MA: Academic Publisher, 132163. Luckow M, Miller JT, Jobson R, Murphy DJ. 2004. Generic relationships in the Mimosoideae (Fabaceae). Botany 2004, Alpine Diversity: Adapted to the Peaks, Salt Lake City, Utah, USA. Session 19 (Colloquium) Rupert Barneby and his legume legacy. Abstract 114, 29. Magallon-Puebla S, Ceballos-Ferriz SRS. 1994. Fossil legume fruits from tertiary strata of Puebla, Mexico. Canadian Journal of Botany 72: 10271038. Malcomber S. 2002. Phylogeny of Gaertnera Lam. (Rubiaceae) based on multiple DNA markers: evidence of a rapid radiation in a widespread, morphologically diverse genus. Evolution 56: 4257. Miller JT, Bayer RJ. 2001. Molecular phylgenetics of Acacia (Fabaceae: Mimosoideae) based on the chloroplast matK conding sequence and anking trnK intron spacer regions. American Journal of Botany 88: 697705. Moore MJ, Jansen RK. 2006. Molecular evidence for the age, origin, and evolutionary history of the American desert

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640

636

S. A. CATALANO ET AL.
ed. Nursery technology of forest tree species of arid and semiarid regions. New Delhi: Winrock-Oxford & IBH Publishing Co. PVT. Ltd., 187198. Saidman BO, Vilardi JC, Pocov MI, Acreche N. 1996. Isozyme studies in Argentine species of the Section Strombocarpa, Genus Prosopis (Leguminosae). Journal of Genetics 75: 139149. Saidman BO, Vilardi JC, Montoya S, Dieguez MJ, Hopp HE. 1998. Molecular markers: a tool for the understanding of the relationships among species of Prosopis (Leguminosae, Mimosoideae). In: Puri S, ed. Tree improvement: applied research and technology transfer. Eneld, NH: Science Publishers, Inc., 324. Sanderson MJ. 2002. Estimating absolute rates of molecular evolution and divergence times: a penalized likelihood approach. Molecular Biology and Evolution 19: 101109. Sanderson MJ. 2003. r8s: inferring absolute rates of molecular evolution and divergence times in the absence of a molecular clock. Bioinformatics 19: 301302. Schrire BD, Lavin M, Lewis GP. 2005. Global distribution patterns of the Leguminosae: insights from recent phylogenies. In: Friis I, Balslev H, eds. Plant diversity and complexity patterns: local, regional and global dimensions. Biologiske Skrifter 55: 375422. Simmons MP, Ochoterena H. 2000. Gaps as characters in sequence-based phylogenetic analyses. Systematic Biology 49: 369381. Strand AE, Leebens-Mack J, Milligan BG. 1997. Nuclear DNA-based markers for plant evolutionary biology. Molecular Ecology 6: 113118. Swofford DL. 2002. PAUP* phylogenetic analysis using parsimony (*and other methods), Version 4. Sunderland, MA: Sinauer Associates. Taberlet P, Gielly L, Pautou G, Bouvet J. 1991. Universal primers for amplication of three non-coding regions of chloroplast DNA. Plant Molecular Biology 17: 11051109. Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins DG. 1997. The CLUSTALwindows interface: exible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Research 25: 4876 4882. UNEP. 1992. World atlas of desertication. London: Edward Arnold. Verboom GA, Linder HP, Stock WD. 2003. Phylogenetics of the grass genus Ehrharta: evidence for radiation in the summer-arid zone of the South African cape. Evolution 57: 10081021. Verga AR. 1995. Genetische untersuchungen an Prosopis chilensis und Prosopis exuosa (Mimosaceae) im trockenen Chaco Argentiniens. Gttingen: Universitt Gttingen. Volkheimer W. 1971. Aspectos paleoclimticos del Terciario argentino. Revista del Museo Argentino de Ciencias Naturales Bernardino Rivadavia 1: 243262. Von Hagen KB, Kadereit JW. 2001. The phylogeny of Gentianella (Gentianaceae) and its colonization of the southern hemisphere as revealed by nuclear and chloroplast DNA sequence variation. Organisms Diversity and Evolution 1: 6179.

plant genus Tiquilia (Boraginaceae). Molecular Phylogenetics and Evolution 39: 668687. Naranjo CA, Poggio L, Enus Zeiger S. 1984. Phenol chromatography, morphology and cytogenetics in three species and natural hybrids of Prosopis (LeguminosaeMimosoideae). Plant Systematic and Evolution 144: 257 276. Nee S, Holmes EC, Rambaut A, Harvey PH. 1995. Inferring population history from molecular phylogenies. Philosophical Transactions of the Royal Society of London, Series B 349: 2531. Palacios RA, Bravo LD. 1981. Hibridacin natural en Prosopis (Leguminosae) en la region chaquea argentina. Evidencias morfologicas y cromatogrcas. Darwiniana 23: 335. Pascual R, Ortiz-Jaureguizar E, Prado JL. 1996. Land mammals: paradigm of Cenozoic South American geobiotic evolution. In: Arratia G, ed. Contribution of Southern South America to Vertebrate Paleontology. Munich: Mnchner Geowissenschaftliche Abhandlungen (A), 265319. Pasiecznik NM, Felker P, Harris PJC, Harsh LN, Cruz G, Tewari JC, Cadoret K, Maldonado LJ. 2001. The prosopis juliora-prosopis pallida complex: a monograph. Coventry: HDRA. Piel KM. 1971. Palynolgy of Oligocene sediments from British Columbia. Canadian Journal of Botany 49: 18851920. Posada D, Crandall KA. 1998. Modeltest: testing the model of DNA substitution. Bioinformatics 14: 817818. Rambaut AP. 2002. Phylogen, Version 1.0. Oxford: Department of Zoology, University of Oxford. Ramrez L, De La Vega A, Razkin N, Luna V, Harris PJC. 1999. Analysis of the relationships between species of the genus Prosopis revealed by the use of molecular markers. Agronomie 19: 3143. Richardson JE, Pennington RT, Pennington TD, Hollingsworth PM. 2001. Rapid diversication of a species-rich genus of Neotropical rain forest trees. Science 293: 22422245. Riddle BR, Hafner DJ. 2006. A step-wise approach to integrating phylogeographic and phylogenetic biogeographic perspectives on the history of a core North American warm deserts biota. Journal of Arid Environments 66: 435461. Roig FA. 1993. Informe nacional para la seleccin de germoplasma en especies de Prosopis en la Repblica Argentina. Mendoza, Argentina: IADIZA-CRICYT, 137. Ronquist FR, Huelsenbeck JP. 2005. Mr Bayes 3. Bayesian phylogenetics inference under mixed models. Bioinformatics 19: 15721574. Ross JH. 1975. Flora of Southern Africa. Vol. 16, Part I. Mimosoideae, Fabaceae. Pretoria: Deptartment of Agriculture. Rzedowski J. 1988. Analisis de la distribucin geogrca del complejo Prosopis (Leguminosae, Mimosoideae) en Norteamerica. Acta Botanica Mexicana 3: 719. Saidman B, Vilardi J. 1987. Analysis of the genetic similarities among seven species of Prosopis (Leguminoseae: Mimosoideae). Theoretical and Applied Genetics 75: 109 116. Saidman BO, Vilardi JC. 1993. Genetic variability and germplasm conservation in the genus Prosopis. In: Puri S,

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640

PROSOPIS PHYLOGENY AND EVOLUTION

Wang A, Yang M, Liu J. 2005. Molecular phylogeny, recent radiation and evolution of gross morphology of the rhubarb genus Rheum (Polygonaceae) inferred from chloroplast DNA trnL-F sequences. Annals of Botany 96: 489498. Xiang Q, Manchester SR, Thomas DT, Zhang W, Fan C. 2005. Phylogeny, biogeography, and molecular dating of

637

cornelian cherries (Cornus, Cornaceae): traking tertiary plant migration. Evolution 59: 16581700. Yule GU. 1924. A mathematical theory of evolution based on the conclusions of Dr. J. C. Willis. Philosophical Transactions of the Royal Society of London, Series B 213: 21 87.

SUPPLEMENTARY MATERIAL
The following material is available for this article online: Figure S1. Strict consensus of 4680 most parsimonious trees obtained in the ve-marker analysis (trnS-psbC, G3pdh, NIA, trnL-trnF, trnK-matK). Table S1. Sequences downloaded from GenBank. Table S2. Maximum and minimum humidity index for species included in the ancestral climate estimation analysis. This material is available as part of the online article from: http://www.blackwell-synergy.com/doi/abs/10.1111/j.1095-8312.2007.00907.x (This link will take you to the article abstract). Please note: Blackwell Publishing are not responsible for the content or functionality of any supplementary materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640

APPENDIX 1

638

List of species sequenced in the present study, indicating the Section and Series to which they belong, voucher specimen, the natural distribution and accession numbers

Species trnS-psbC EF165309 EF165297 EF165253 EF165269 EF165292 EF165291 NIA trnL-trnF DANIDA N 01089/82 HDRA N63 AF EF165209 AF EF165221

Section Series

Voucher specimen/seed bank accesion

Natural distribution

GenBank Accesion numbers G3pdh

trnK/matK

Prosopis

S. A. CATALANO ET AL.

Anonychium

EF165287/ EF165248 EF165286/ EF165251

Hunziker & Gamerro 11359 (SI) Saidman 36 (BAFC) SA, NA SA EF165244 EF165330 EF165237 EF165333 Hunziker 9733 (SI)

SA

EF165241

EF165327

EF165281

Strombocarpa Strombocarpae Strombocarpa Strombocarpae Strombocarpa Strombocarpae Evans N 15 (USDA-USA) NA SA SA SA SA EF165223 EF165242 EF165212 EF165240 EF165234 Hunziker et al. 10451 (SI) DANIDA N 01215/83 NA EF165236 EF165323 EF165322 EF165311 EF165328 EF165300 EF165326

EF165278 EF165261 EF165280 EF165293 EF165290

EF165289/ EF165250 EF165288/ EF165249

Strombocarpa Strombocarpae Strombocarpa Strombocarpae Strombocarpa Cavenicarpae Strombocarpa Cavenicarpae Monilicarpa BNG Prosopis N 15-88. Saidman & Vilardi 961 (BAFC) SA

Algarobia Sericanthae

Prosopis cineraria (L.) Druce Prosopis africana (Guill., Perr., & Rich.) Taubert Prosopis strombulifera (Lam.) Bentham Prosopis reptans Bentham Prosopis torquata (Cavanilles ex Lagasca) Prosopis pubescens Bentham Prosopis palmeri S. Watson Prosopis ferox Grisebach Prosopis tamarugo F. Philippi Prosopis argentina Burkart Prosopis sericantha Gilles ex Hooker & Arnott Prosopis kuntzei Harms EF165230 EF165231 SA SA EF165239 EF165245 EF165246 SA SA EF165222 EF165238 EF165318 EF165319 EF165325 EF165331 EF165332 EF165310 EF165324 EF165276 EF165277 EF165257 EF165283 EF165284 EF165270 EF165279

Algarobia Sericanthae

Prosopis ruscifolia Grisebach Prosopis vinalillo Stuckert

Algarobia Ruscifoliae Algarobia Ruscifoliae

Saidman & Vilardi 516 (BAFC); Saidman & Vilardi 521 (BAFC) Saidman & Vilardi 420 (BAFC) Saidman & Vilardi 505 (BAFC); Saidman & Vilardi 511 (BAFC) Catalano 33 (BAFC) Catalano 13 (BAFC)

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640

Prosopis denudans Bentham Prosopis ruizleali Burkart

Algarobia Denudantes Algarobia Denudantes

Catalano 8 (BAFC) Catalano 1 (BAFC) SA EF165227 EF165315 EF165275

Algarobia Denudantes Algarobia Humiles SA SA NA NA SA EF165219 EF165220 EF165307 EF165308 EF165268 EF165285 EF165243 EF165329 EF165282 EF165213 EF165301 EF165262 Catalano 18 (BAFC); Catalano 20 (BAFC) Saidman et al. 526 (BAFC) EF165216 EF165217 EF165208 EF165304 EF165305 EF165296 EF165265 EF165266 EF165256

SA

EF165218

EF165306

EF165267

Algarobia Pallidae Algarobia Pallidae Algarobia Pallidae Algarobia Pallidae Algarobia Chilenses NA,CA,SA SA EF165228 EF165229 EF165232 EF165233 EF165214 EF165215 EF165224 EF165225 EF165226 EF165211 EF165210 EF165235 EF165302 EF165303 EF165312 EF165313 EF165314 EF165299 EF165298 EF165295 EF165316 EF165317 EF165320 EF165321 EF165258 EF165274 EF165259 EF165260 EF165263 EF165264 EF165271 EF165272 EF165273 EF165255 EF165254

Algarobia Chilenses Algarobia Chilenses SA

Prosopis castellanosii Burkart Prosopis humilis Gilles ex Hooker & Arnott Prosopis campestris Grisebach Prosopis affinis Sprengel Prosopis articulata S. Watson Prosopis tamaulipana Burkart Prosopis chilensis (Molina) Struntz emend. Burkart Prosopis juliora (Swartz) DC Prosopis nigra (Grisebach) Hieronymus Prosopis caldenia Burkart SA

Algarobia Chilenses

Prosopis exuosa DC

Algarobia Chilenses NA SA SA SA

Saidman & Vilardi 911 (BAFC); Saidman & Vilardi 922 (BAFC) Hunziker 10039 (SI); Hunziker 10040 (SI) Saidman & Vilardi 435 (BAFC); Saidman 315 (BAFC) Saidman & Vilardi 350 (BAFC); Saidman & Vilardi 138 (BAFC) Saidman & Vilardi 302 (BAFC); Catalano 10 (BAFC) Evans N 5 (USDA-USA) Catalano 34 (BAFC) Saidman & Vilardi 413 (BAFC) Peter 244 (LP)

/EF165252

Algarobia Chilenses Algarobia Chilenses Algarobia Chilenses NA

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640

NA

AF SA SA

EF585492 EF165207 EF165247

EF585493 EF165294 Supp. data

EF165206

Supp. data Supp. data Supp. data

Supp. data Supp. data Supp. data

PROSOPIS PHYLOGENY AND EVOLUTION

Prosopis glandulosa Torrey Prosopis alpataco R.A. Philippi Prosopis alba Grisebach Prosopidastrum angusticarpum Palacios & Hoc Xerocladia viridiramis Taub. Acacia caven (Mol.) Molina Lotus japonicus L. Germishuizen 8244 (PRE) Saidman et al. 789 (BAFC)

NA

NA

BNG Prosopis, Banco Nacional de Germoplasma de Prosopis (Cordoba, Argentina); DANIDA, Danida Forest Seed Centre, Denmark; GRS-USDA-USA, United States Department of Agriculture; HDRA, Henry Doubleday Research Association, United Kindom; SI, Darwinion Herbarium, San Isidro, Argentina; BAFC, Herbarium of Facultad de Ciencias Exactas y Naturales UBA, Buenos Aires, Argentina; AF, Africa; AS, Asia; CA, Central America; NA, North America; SA, South America; Supp. data, accession number included in Supplementary material (Table S1).

639

640

S. A. CATALANO ET AL.

APPENDIX 2
SEARCH
STRATEGIES IN PARSIMONY AND

BAYESIAN

ANALYSES

In three-marker analysis, parsimony searches involved 100 Random Addition Sequences (RAS) followed by ten cycles of Tree Drifting. In the twomarker and ve-marker analyses, the strategy involved 200 RAS followed by Tree Drifting and Random Sectorial Searches (Goloboff, 1999). These two algorithms intend to overcome the problem of local optima during tree searches. In Tree Drifting, suboptimal solutions are accepted with a certain probability in different moments of the swapping whereas, in Sectorial Searches, reduced datasets are created for different parts of the tree which are analysed by tree-bisection-reconnection (TBR). Then, the best tree for each sector is replaced in the original tree. In all cases, trees found using these strategies were used as starting trees for a nal round of TBR keeping a maximum of 10 000 trees. Constrained searches forcing groups not found in the most parsimonious trees were run with the same settings.

Support was evaluated with jackkning procedure with an independent character removal probability of 0.36. For each jackkning replicate, search strategy was the same than previously described except that the number of RAS was decreased to 10. Absolute Bremer Support (Bremer, 1994) was calculated saving up to 6000 suboptimal trees in successive steps up to trees 12 steps longer than the optimal length. Bayesian phylogenetic analyses were performed with the Metropolis-coupled Markov chain Monte Carlo algorithm implemented in MrBayes, version 3.1.2 (Ronquist & Huelsenbeck, 2005). The optimal model was selected with Modeltest (Posada & Crandall, 1998) and the values of each parameter were determined during each run. The analysis involved two independent runs of 2 106 generations sampled every 100 generations after a burnin period of 0.50 106 generations. Each run involved three heated chains. Convergence was assessed evaluating likelihood plots and standard deviation of the split frequencies.

2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 621640